State of the State: Generative AI

Generative AI is undoubtedly the hottest topic right now, and not just within marketing circles; it’s dominating legislative, scientific, and cultural discussions at large. But this space is also in a state of constant evolution, which means that keeping up with all its twists and turns is no small task.

Enter Collective Measures’ new generative AI POV series, where each month we will distill the state of the state, outlining critical trends, advancements, conversation topics, and — of course — related marketing takeaways. This POV unpacks:

The ever-growing list of new generative AI technology announcements from media/technology platforms and brands alike
Frequently cited AI concerns that are dominating conversations at all levels: consumer, brand, government, and society at large
Potential marketing applications for generative AI

Here’s what we know as of April 1, 2023.

New technologies: it’s an arms race

While generative AI is currently front and center, it is not new technology — both image and text-based generative AI have been around for years. That said, use cases have been somewhat limited (think chatbots on websites), leveraged almost exclusively by developers and marketers, and their accuracy was constantly in question. ChatGPT swiftly changed the game, largely because it is one of the first consumer-friendly generative AI tools. It’s also free, easily accessible, and when applied to the right use cases, strikingly accurate.

ChatGPT is the match that ignited the cultural fire that’s currently driving both consumer interest and tech development. As a result, technology platforms and brands alike are now locked in an arms race to develop and launch their own versions of generative AI tools, whether by fast tracking pre-existing projects or quickly kicking off new development efforts. Here’s a quick snapshot of the ever-evolving landscape (including a helpful repository of all the AI tech currently available):

Platforms:

Media and technology platforms coming out with their own AI:

Google’s chatbot, BARD, is now open to the public for testing
Mozilla announced a $30M investment in ethical AI
OpenAI and Microsoft are introducing GPT-4, a new, more advanced AI model
Snap is launching MyAI for Snapchat+ subscribers, which will incorporate GPT technology into Snapchat
Bing’s AI-powered Image Creator tool will help Bing Chat users create images using words to describe the desired picture
Meta’s forthcoming AI language model (LLaMA), which was reportedly leaked early, will be opening to the AI community; it will not be a chatbot open to the public
Adobe’s product, Sensei Gen AI, will help marketers create text-based message variations for key consumer touchpoints. It also announced Firefly, a family of generative AI models that will be integrated into Adobe Experience Cloud

Brands:

Simultaneously, brands are working to infuse AI into their own businesses, whether by coming up with an AI model of their own or working with developers to integrate various generative AI models into their own technology. Examples here are growing by the day, but think about Zoom’s quest to help attendees catch up on meetings they were late to or Duolingo’s new Duolingo Max, a learning experience powered by GPT-4.

A related and interesting development from OpenAI is ChatGPT plugins. This is in the early stages for testing, but the end goal is to allow website owners to develop, utilize, and integrate ChatGPT plugins into their site. For example, think of a search function built into a website that could answer specific questions about its products and services driven by ChatGPT via API. Some companies (including Expedia, Instacart, KAYAK, and more) have been very quick to integrate this; however, many of these early movers are likely building on/increasing the sophistication of existing tech or functionality. We anticipate seeing many more ChatGPT plugin use cases in coming months.

Key takeaways:

Democratizing AI. While AI has been part of marketing technology for some time, it’s never been so visible to the general public; this is largely what’s fueling the current interest. Because these tools are publicly accessible — and increasingly prevalent across platforms and brand websites alike — we expect that technology will expand and innovate faster.
Higher tolerance for error. With so many companies rushing to market with new technologies and innovations, success will likely be measured by speed instead of accuracy. This means that new products will almost certainly have bugs, which users will increasingly come to not only tolerate, but expect. However, this also places even greater importance on scrutinizing AI-generated content for accuracy.

Concerns about AI

Not surprisingly, there are just as many conversations around AI concerns as there are about AI advancements. And rightfully so — we are entering uncharted territory, and the long-term implications on culture and society are unknown.

So, while we see incredible opportunity to explore generative AI in its many forms, we also believe it’s important to go in with eyes wide open. A couple of the biggest stories we’re following include:

General apprehension around AI tech. Hundreds of scientists, AI experts, and tech professionals recently signed a petition calling for a pause in the development and advancement of powerful AI technologies to assess the risks and consequences of these technologies on society.
The impact on the job market. This has been covered extensively in the news given the potential changes generative AI could bring to traditional workstreams and processes. Some sources have projected significant job losses across major economies.
Security fears from governments and brands alike. Specific examples include the EU sounding the alarm about potential criminal use of ChatGPT and concerns that employees may feed brand-sensitive or even proprietary information into the tool.
The potential for misinformation. There’s endless conversation around this topic, and for good reason — it’s too easy to accept the information ChatGPT spits out as fact. This is especially worrisome given a recent NewsGuard analysis that discovered GPT-4 advanced prominent false narratives not only more frequently, but also more persuasively, than previous iterations of the tool. When considering potential GPT use cases, this is a critical consideration.
Copyright and plagiarism concerns. AI learns from various inputs and then uses them to create net-new copy. The problem is that sometimes that copy is parroted — or worse, taken word for word — from the sources it learns from, which means that anyone using AI to generate content could accidentally publish plagiarized copy. Online tools that check for plagiarism (for example, Grammarly) can help avoid this outcome.
Leveling the playfield within the AI landscape. In response to concern over big tech companies dominating the AI space — and therefore stifling innovation — the FTC announced it is monitoring the competitive AI field. The FTC also stated a commitment to ensuring companies are not “overstating or overselling” their AI capabilities and integrations. While today’s playing field feels relatively even given the number of companies jumping into AI, the FTC’s involvement showcases the level of importance being placed on this landscape.

Key takeaways:

Uncharted waters. It’s impossible to know exactly where AI technologies are headed. There will certainly be challenges for marketers and businesses to overcome, as well as opportunities to take advantage of AI’s benefits. The successes and failures of social media over the last 15 years may be the best guide available. Hopefully, we can learn from our mistakes to develop and utilize AI in positive ways, with thoughtful oversight.
Approach with caution. With little oversight or legal precedent, those looking to leverage AI will need to approach the technologies with caution. Considering all possible outcomes of AI integration will prepare business leaders for challenges they may face regarding security, misinformation, and workforce impact.

Considering use cases for AI

Despite all the concerns outlined above, generative AI is here to stay. Using Chatbots to assist with customer service is a quick win, but there are essentially unlimited use cases for AI across all facets of business, from logistics and the supply chain to marketing and more. Within the walls of CM, here are a few things we are currently evaluating:

Comparing Bing Chat (ChatGPT) to Google’s BARD. We have members of Collective Measures enrolled in beta programs across both Bing’s and Google’s respective generative AI products. As these programs continue to take shape, we’re documenting comparative experiences and features between the two. We’ll share findings in the coming weeks.
Identifying testing opportunities using emerging tools. While opportunities here differ by client, we’re keeping a pulse on this landscape. One example? Typeface, a tool aimed at helping marketers create image assets and copy for easier social campaign development and testing. It also offers protection from plagiarism/copyright infringement and comes with some serious backing in terms of the money and the talent behind it.

Discussing the pros and cons of generative AI for SEO and content marketing. There’s no doubt that generative AI can impact content marketing efforts. The real questions are (1) what role it should play in the process and (2) what is the appropriate — and necessary — balance between human and machine? We’re continuing to finetune this process but have found success incorporating ChatGPT as another research tool that our SEOs use as an input (not a replacement) for content ideation and generation.

Want to learn more? Contact us at info@collectivemeasures.com